The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill provides mechanisms to ingest and process untrusted data from external SIP signaling, creating a potential surface for indirect prompt injection.\n- Ingestion points: Data enters through SipDigestAuth.parseAuthorizationHeader, SipRegistrar.handleRegister (processing headers), and SipTlsServer.handleData (processing raw socket data).\n- Boundary markers: The code includes regex-based validation within the SipSecurityValidator class to verify and delimit input formats.\n- Capability inventory: The skill utilizes fs.readFileSync for certificate management and tls for network communication, both of which are reachable by code paths processing external data.\n- Sanitization: Sanitization is implemented via validation methods like validateUri and validateSdp, which filter for injection patterns and dangerous characters.

sip-authentication-security